> ## Documentation Index
> Fetch the complete documentation index at: https://docs.heygarth.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# AI Coding Tool Settings

> Configure the cost-saving assumptions and waste/score thresholds G360 uses to analyze your AI coding tool usage.

The **AI Coding Tool Settings** tab is where you configure the analytics assumptions G360 uses when analyzing usage of your AI coding tools — things like expected savings from cost-optimization features, and the thresholds used to flag waste. This tab lives under **Administration → Settings → AI Coding Tool Settings**.

<Note>
  This is a different tab from the **AI Coding Tool Settings** sub-tab found under **ROI Settings**. That sub-tab configures per-seat cost and hours-saved figures for ROI; this tab configures the deeper analytics parameters (downgrade savings, waste thresholds, score weights) behind those numbers.
</Note>

Underneath this tab, a row of tool tabs lets you switch between AI coding tools — currently **Claude Settings** is available.

## Claude Settings

<Frame>
  <img src="https://mintlify.s3.us-west-1.amazonaws.com/na-f445b234/images/claude-code-analytics-settings.png" alt="AI Coding Tool Settings - Claude Settings" />
</Frame>

**Claude Code Analytics Settings** shows when it was last updated (for example, "11 Aug 2026, 07:45") and is organized into several sections.

### Model Downgrade

Controls when G360 considers a request eligible for automatic model downgrading (switching to a cheaper model) and how much that's expected to save.

| Field               | Details                                                                                    |
| ------------------- | ------------------------------------------------------------------------------------------ |
| **Token Threshold** | The minimum number of tokens a request must use before a downgrade is triggered.           |
| **Saving Rate**     | The expected cost reduction (%) from downgrading a request that meets the threshold above. |

### Savings Rates

Configures the expected cost savings from two specific cost-optimization behaviors.

| Field                    | Details                                                                                                   |
| ------------------------ | --------------------------------------------------------------------------------------------------------- |
| **Caching Saving Rate**  | The expected cost reduction (%) from prompt caching.                                                      |
| **Coaching Saving Rate** | The expected cost reduction (%) from usage coaching (guiding users toward more efficient usage patterns). |

### High Waste Thresholds

Sets the cutoff percentiles G360 uses to flag usage as high-waste.

| Field                      | Details                                                                                                         |
| -------------------------- | --------------------------------------------------------------------------------------------------------------- |
| **Token Percentile**       | The percentile cutoff for flagging high-waste token usage — usage above this percentile is considered wasteful. |
| **Completions Percentile** | The percentile cutoff for flagging high-waste completions.                                                      |

### Score Weights

Configures how much weight different usage signals carry when G360 calculates an overall efficiency/usage score. Fields visible include:

| Field                       | Details                                                                     |
| --------------------------- | --------------------------------------------------------------------------- |
| **Completions per Session** | The weight (%) given to completions-per-session when calculating the score. |
| **Token Efficiency**        | The weight (%) given to token efficiency.                                   |
| **Acceptance Rate**         | The weight (%) given to acceptance rate.                                    |

<Note>
  The Score Weights section continues below what's captured here — additional weight fields may appear further down the page. Scroll down within this section to see and configure all of them.
</Note>

Once you've made your changes, click **Save Settings** in the top right to apply them, or **Reset** to revert all fields back to their last saved values.